Long-term temperature stations in mountainous terrain: confounding processes in interpreting high-elevation climate trends

Mountains are spatially complex and sparsely sampled. Temperatures patterns differ diurnally, synoptically, and seasonally and do not always increase linearly with elevation. However, spatial maps of temperature are imperative to understand spatial patterns of ecology, snowmelt, climate change, and frost. Hundreds of self-recording temperature sensors, such as the Onset Hobo and the Maxim iButton, have been deployed in Yosemite National Park, California; Niwot Ridge and Rocky Mountain National Park, Colorado; the Eastern Pyrenees, France; and Mt. Rainier, Washington over the past several years. The resulting temperature data, combined with empirical orthogonal functions (EOFs), can be used to identify the dominant spatial temperature patterns within each study area and how they vary in time. The spatial patterns of temperature are correlated with topography, such as windward-slope, lee-slope, valley, or ridge. Some patterns, such as nocturnal drainage and cold air pools, appear in all the study areas. Here we demonstrate an automated algorithm for using a digital elevation model to map where cold air pools and test it against data from the study sites listed above. These maps are then used to interpret site-specific variations in long-term temperature trends.